The authors have declared that no competing interests exist.
Conceived and designed the experiments: LC CL PM. Performed the experiments: LC LG VD PM. Analyzed the data: LC CL PM. Wrote the paper: LC PM.
GFP-tagged proteins are used extensively as biosensors for protein localization and function, but the GFP moiety can interfere with protein properties. An alternative is to indirectly label proteins using intracellular recombinant antibodies (scFvs), but most antibody fragments are insoluble in the reducing environment of the cytosol. From a synthetic hyperstable human scFv library we isolated an anti-tubulin scFv, 2G4, which is soluble in mammalian cells when expressed as a GFP-fusion protein. Here we report the use of this GFP-tagged scFv to label microtubules in fixed and living cells. We found that 2G4-GFP localized uniformly along microtubules and did not disrupt binding of EB1, a protein that binds microtubule ends and serves as a platform for binding by a complex of proteins regulating MT polymerization. TOGp and CLIP-170 also bound microtubule ends in cells expressing 2G4-GFP. Microtubule dynamic instability, measured by tracking 2G4-GFP labeled microtubules, was nearly identical to that measured in cells expressing GFP-α-tubulin. Fluorescence recovery after photobleaching demonstrated that 2G4-GFP turns over rapidly on microtubules, similar to the turnover rates of fluorescently tagged microtubule-associated proteins. These data indicate that 2G4-GFP binds relatively weakly to microtubules, and this conclusion was confirmed in vitro. Purified 2G4 partially co-pelleted with microtubules, but a significant fraction remained in the soluble fraction, while a second anti-tubulin scFv, 2F12, was almost completely co-pelleted with microtubules. In cells, 2G4-GFP localized to most microtubules, but did not co-localize with those composed of detyrosinated α-tubulin, a post-translational modification associated with non-dynamic, more stable microtubules. Immunoblots probing bacterially expressed tubulins confirmed that 2G4 recognized α-tubulin and required tubulin’s C-terminal tyrosine residue for binding. Thus, a recombinant antibody with weak affinity for its substrate can be used as a specific intracellular biosensor that can differentiate between unmodified and post-translationally modified forms of a protein.
Localization of proteins within cells typically relies on either expression of fluorescently tagged proteins or by labeling proteins with specific antibodies. While each method is powerful, each has limitations. Expression of fluorescently tagged proteins allows dynamic processes to be followed in living cells, but often represents over-expression of the protein of interest and the large size of the fluorescent protein can interfere with the function. For example, gap junctions assembled from connexin-43 tagged at its C-terminus are much larger than gap junction plaques assembled from untagged connexin-43; the larger plaque size is due to the loss of binding between GFP-tagged connexin-43 and zonula occludens-1
The most widely used recombinant antibody format is expression of a single chain polypeptide encompassing the heavy and light chain variable regions from immunoglobin (termed single chain variable region, scFv). Phage display is used to screen these libraries and has allowed selection of scFv specific for protein conformation (e.g., tubulin bound to GTP
Heterodimers of α and ß tubulin polymerize to form microtubules, polymers responsible for a number of cellular functions including chromosome movement in mitosis, vesicle transport, intracellular organization and as scaffolds for components of signal transduction pathways
Here we report the intracellular expression of a GFP tagged recombinant scFv, 2G4-GFP, that is soluble in the cytoplasm of cells and co-localizes with microtubules without disrupting microtubule polymerization dynamics. Fluorescence recovery after photobleaching and in vitro microtubule co-pelleting experiments indicate that 2G4-GFP binds microtubules relatively weakly, with an affinity for microtubules similar to several microtubule-associated proteins (MAPs). This scFv specifically recognizes tyrosinated, but not detyrosinated, α-tubulin. We suggest that visual screens of GFP-tagged scFvs could be used to isolate scFv recognizing specific post-translationally modified states of a protein of interest, and that scFvs showing a localization pattern in living cells similar to antibody localization in fixed cells are likely to bind with sufficiently weak affinity to mark a protein without compromising its function. Such scFv could be combined with protein conformation-sensitive scFv
Hela, LLCPK and MEF cells, obtained from ATCC, were maintained as described previously
Cells were fixed in −20°C methanol/EDTA and stained as described previously
For experiments exploring scFv binding to purified tubulin, scFv’s tagged with 6x His were expressed in E. coli and purified as described previously
Genes encoding human TUBA1A and TUBB were inserted in frame at the 3′ extremity of the GST gene in the pGEX-4T1 vector. Deletion of the C-terminal tyrosine of the TUB1A1 gene was performed by PCR. All inserts were fully sequenced. TG1 cells containing plasmids were grown in LB/ampicillin medium to an OD 600nm of 0.4, then 1 mM of IPTG was added and the culture continued for 3 h at 37°C with shaking. Cells were pelleted and resuspended in the same volume of Laemmli buffer (0.01% bromophenol blue, 5% β-mercaptoethanol, 2% SDS, 10% glycerol, 62.5 mM Tris-HCl pH 6.8), and boiled. Proteins were separated on a 10% SDS-PAGE gel and transferred on a nitrocellulose membrane. The membrane was saturated in PBSTM (PBS, Tween 0.1%, Non-fat milk 2%) for 1 h, incubated with purified 2G4 (10 µg/ml) for 2 h in the same buffer. Bound scFv was revealed using an HRP-coupled anti-His-tag antibody (HIS-1, monoclonal; Sigma H1029), which recognizes the His tag on bacterially expressed scFv. A rabbit anti-GST polyclonal antibody was used to confirm equal loading of GST-tubulin fusion proteins. Expression of the detyrosinated mutant was confirmed using a rabbit antibody specific for detyrosinated a-tubulin
Soluble porcine brain extracts were separated on a 10% SDS-PAGE gel (1 µg/lane) and then transferred to a nitrocellulose membrane. Membranes were saturated in PBSTM for 1 h and incubated with 13R4 (10 µg/ml), 2G4 (10 µg/ml), 2F12 (2 µg/ml), a rabbit polyclonal serum raised against purified tubulin (a kind gift of Ned Lamb), or a monoclonal anti-tubulin (clone Tub2.1; Sigma T4026). Bound antibodies were revealed with a suitable HRP-coupled secondary antibody (anti-His-tag Sigma H1029 for the 3 scFv’s) and enhanced chemiluminescence detected with a Hyperfilm.
Pig brains were obtained from the Abattoir of Aves (France) under supervision of local employees and the staff veterinarian. Tubulin was isolated from porcine brains by a combination of two cycles of microtubule assembly/disassembly, followed by phosphocellulose chromatography
For observations of living cells expressing fluorescently tagged proteins, cells were plated onto either glass bottom dishes (MatTek Corp., Ashland, MA) or 30 mm round coverslips, and transferred to high glucose DMEM lacking phenol red and supplemented with 10 mM HEPES pH 7.3, 10% fetal bovine serum, 1 mM sodium pyruvate just prior to imaging. For photobleaching experiments, 30 mm round coverslips were mounted into a POCmini chamber (PeCon GmbH, Erbach, Germany). The POCmini chamber was maintained at 37°C on a Zeiss LSM confocal microscope stage through use of stage and objective heaters (PeCon GmbH, Erbach, Germany). Cells on glass bottom dishes were imaged on a swept field confocal system (described below); temperature was maintained at 37°C by an environmental chamber (InVivo Scientific, St. Louis, MO).
Photobleaching was performed using a Zeiss LSM510META confocal microscope as described previously
Microtubule dynamics, as marked by 2G4-GFP or GFP-α-tubulin, were imaged using a swept field confocal system (Prairie Technologies, Middleton, WI) mounted to a Nikon Eclipse Ti inverted microscope equipped with a 100X/1.4 NA planapo objective. A 488 nm solid-state sapphire laser (30 mW, used at 5 - 10% transmission) was used to excite GFP. Images were collected using a 35 mm slit aperture and an exposure time of 900 ms. Images were projected through a 1.5X optivar to a Photometrics QuantEM 512SC camera. Image acquisition was controlled by NIS Elements AR software version 3.0 (Nikon Imaging, Melville NY). Images were exported as TIFF files and imported into MetaView imaging software (Molecular Devices, Downingtown, PA) for analysis of microtubule length changes over time
The rate of fluorescence recovery after photobleaching was calculated from integrated fluorescence intensity measurements within either the bleached region or along a line on single microtubules. Zeiss LSM software was used to measure the integrated fluorescence intensity. Data were exported to an Excel spreadsheet for further calculations. Fluorescence intensities were normalized by setting the average pre-bleach fluorescence to 100. Fluorescence intensity within an unbleached area (or from a line along an unbleached microtubule) was used to correct for photobleaching caused by image acquisition. Plots are shown as normalized, corrected fluorescence over time. Fluorescence recovery rate was calculated from an exponential function, as described previously
All quantitative data are presented as mean ± SD. Statistical significance was determined using unpaired t-tests.
Previously we reported that an anti-tubulin scFv, 2G4, can be expressed as a GFP fusion protein in human cell lines
(A) Diagram outlining the recombinant antibody. The combined mw of the VH and VL regions approximately equals that of EGFP. (B) 2G4-GFP localizes to linear filaments in either living cells or in cells fixed in −20°C methanol. Images shown are from the edges of two different LLCPK cells (scale bar = 5 µm). (C) 2G4-GFP localizes to the majority of microtubules. LLCPKs were transfected with plasmid encoding 2G4-GFP and fixed 24 h later. Microtubules were stained with an antibody to a-tubulin. Images in the top row show a maximum intensity projection from a Z series (scale bar = 10 µm). The bottom row shows single optical section from the edge of a second LLCPK cell (scale bar = 5 µm).
We confirmed that 2G4-GFP bound to the distal ends of microtubules by co-labeling 2G4-GFP expressing cells with an antibody to end binding protein 1 (EB1), a protein that localizes in a comet-like shape to the distal ∼1 µm of growing microtubules
(A) 2G4-GFP binding extends to the distal ends of microtubules, as marked by EB1. A maximum intensity projection from optical sections through a Hela cell is shown (scale bar = 10 µm). (A′) Single optical sections from bracketed regions of the cell shown in (A) (scale bar = 5 µm). 2G4-GFP binds uniformly along microtubules and extends to the ends of these microtubules. (B) Box plot of EB1 comet lengths at microtubule plus ends. Expression of 2G4-GFP did not change the length of EB1 comets. (C) TOGp, another microtubule plus end binding protein, is localized to microtubules labeled by 2G4-GFP. (D) CLIP-170 was also localized to microtubule plus ends in cells expressing 2G4-GFP. Note that CLIP-170 was localized to only a subset of microtubule ends. We observed a similar pattern in Hela cells expressing GFP- a-tubulin (data not shown). Scale bars in C, D = 2 µm.
Recent data has shown that TOGp binds distal to EB1 at microtubule plus ends
A second anti-tubulin scFv, 2F12-GFP, was present as aggregates when expressed in Hela or LLCPK cells, consistent with our previous results for expression of this scFv in mammalian cells
Microtubule dynamic instability has been well characterized in the LLCPK cell line
(A) Sequential images from the periphery of an LLCPK cell expressing 2G4-GFP. Arrows note length changes of several microtubules. Scale bar = 5 µm. The video sequence is presented in Movie S1. (B) Plots of microtubule length changes over time for three microtubules labeled by 2G4-GFP or by GFP-α-tubulin. Length changes over time were determined from image series as described in Methods (C,D). Microtubule growth (C) and shortening velocities (D) measured by 2G4-GFP or GFP-α-tubulin. Data shown are means ± sd. Additional parameters of dynamic instability are summarized in
Dynamic Instability Parameters | 2G4-GFP | EGFP-Tub | EGFP-Tub |
EGFP-Tub |
Growth Rate (µm/min) | 8.5±4.8 | 10.0±5.0 | 11.5±7.4 | 8.5±5.8 |
Shortening Rate (µm/min) | 12.0±9.4 | 16.9±9.9 | 13.1±8.4 | 11.3±7.9 |
Catastrophe Frequency (s−1) | .027±.004 | .05±.01 | .026±.024 | .053±.003 |
Rescue Frequency (s−1) | .086±0.13 | .106±.024 | .175±.104 | .086±.005 |
Percent Time in growth, shorteningand pause | 63.7, 19.3, 27 | 54.2, 26.6, 19.2 | 15, 11.5, 73.5 | 40.4, 26, 36.5 |
Dynamicity (µm/min) | 8.2±5.6 | 9.34±5.5 | 4±3.5 | 5.3±2.7 |
Parameters of dynamic instability measured using EGFP-α-tubulin (EGFP-Tub) as a tracer, expressed at about 10% of total a-tubulin levels, compared to that measured using 2G4-GFP bound to the external surface of microtubules. Published data sets derived from LLCPK cells expressing EGFP-α-tubulin are also included for comparison. Cells measured in reference
The localization of 2G4-GFP to microtubules, without a measurable impact on microtubule dynamics, stimulated us to ask whether 2G4-GFP bound tightly to microtubules, as one might expect of an antibody, or with a relatively weak affinity, similar to MAPs. As a first estimate of 2G4-GFP binding to microtubules, we measured fluorescence intensity along single microtubules at the distal edges of LLCPK cells and compared these values to the intensity of adjacent microtubule-free cytosol. The intensity of 2G4-GFP bound to microtubules was 1.7 times (±0.4, 29 microtubules in 10 cells) brighter than the adjacent cytoplasm. Given that microtubules take up a small percentage of cell volume, these measurements provide a qualitative indication that most of the 2G4-GFP expressed in a cell is not bound to microtubules, and therefore this scFV may bind only weakly to microtubules.
To estimate 2G4-GFP binding affinity for microtubules within living cells, we used fluorescence recovery after photobleaching (FRAP) to measure 2G4-GFP turnover. We first measured fluorescence recovery within photobleached rectangular shapes positioned near the periphery of cells or in regions closer to the nucleus. Images of cells collected immediately after photobleaching of 2G4-GFP did not yield a distinct photobleached region corresponding to the rectangular area targeted by the laser. Instead, the bleached molecules were distributed over a greater area centered on the rectangular target (
Pre-bleach and post-bleach images are shown for LLCPK cells expressing 2G4-GFP, GFP or GFP-tau as noted. Pre-bleach images were collected immediately prior to photobleaching an area marked by the black rectangular boxes. Post-bleach images were collected immediately after photobleaching. Time in each frame is given in seconds from the start of an imaging experiment. Fluorescence intensity is represented by a rainbow palette from red to blue (see inset at bottom left). For both 2G4-GFP and GFP, the region of dimmed fluorescence has spread beyond that targeted by the bleaching laser. This spread is highlighted by dotted lines in the post-bleach images. The spread of photobleached proteins is likely due to rapid diffusion in and out of the photobleached area. For the GFP pre-bleach image, the nucleus is outlined by a dashed line. Note that fluorescence within the nucleus does not appear to exchange significantly with that in the cytoplasm over the ∼2 s interval between images. In contrast to 2G4-GFP and GFP, GFP-tau photobleaching yields an area of dimmed fluorescence that closely matches the region targeted by the laser. Scale bars for all images are 5 µm.
Based on the first images recorded immediately after photobleaching, it appeared that a large fraction of 2G4-GFP was free to move rapidly within the cytoplasm. To extend this observation, we measured fluorescence recovery within the photobleached rectangular areas. Fluorescence recovery over time reflects both diffusion of soluble 2G4-GFP and 2G4-GFP turnover on microtubules. 2G4-GFP fluorescence quickly recovered after photobleaching (
(A,B) Images of GFP-tau or 2G4-GFP expressing LLCKP cells before (time 0) and after photobleaching rectangular boxes (bleached regions indicated in red) are shown. Time is given in s. (C,D) Typical fluorescence recovery curves for photobleached rectangles in cells expressing GFP-tau or 2G4-GFP. Fluorescence was normalized as described in Methods. (E) Dot plot showing the half times of fluorescence recovery within boxed regions for 2G4-GFP or GFP-tau. Video sequences are available as Movie S2, S3, S4.
For comparison, we expressed and photobleached GFP-tau (isoform with three microtubule-binding repeats). Photobleaching of GFP-tau in rectangular patterns of approximately the same size as those used to photobleach 2G4-GFP resulted in more clearly defined photobleached regions visible in images immediately after photobleaching (
Protein | Half time of fluorescence recovery on microtubules | Reference |
2G4-GFP | 4.5 ± 4.3 s | this study |
GFP-tau | 6.8 ± 4 s | this study |
GFP-tau | 2.3 ± 0.4 s |
|
Ensconsin | 4.3 s |
|
EB1-GFP | 3.6 - 12 s |
|
CLIP-170 | < 1 s |
|
Fibroblast data is shown for GFP-tau
To confirm that the rapid turnover estimated within the photobleached areas included rapid turnover of the fraction of GFP-tagged proteins bound to microtubules, we also measured fluorescence intensity along individual microtubules (
(A) Images shown were recorded immediately after photobleaching. Line scans along individual photobleached microtubules within LLCPK cells expressing either 2G4-GFP or GFP-tau are positioned as indicated. Line scan 1 (red) follows a microtubule region within the photobleached area and line scan 2 (green) follows a microtubule outside the photobleached rectangle. Normalized fluorescence recovery, integrated over each line, is shown below each example. (B) Dot plot showing fluorescence recovery half times.
The cell-based measurements reported above indicated that 2G4-GFP binds microtubules with an affinity similar to MAPs. Co-pelleting of purified 2G4 with in vitro assembled porcine brain microtubules supported the idea that 2G4 binds with relatively weak affinity for microtubules (
Purified porcine brain tubulin was polymerized by addition of GTP and warming to 37°C. After incubation with purified scFv’s, the microtubule fraction was isolated by pelleting through a 40% sucrose cushion and the supernatants and pellets resolved on SDS-PAGE gels. The two anti-tubulin scFv’s, 2F12 and 2G4, co-pelleted with microtubules, but a greater fraction of 2F12 was pelleted compared to 2G4, consistent with comparatively weaker binding of 2G4 to microtubules. 13R4, an anti-ß-galactosidase scFv, did not co-pellet with microtubules, indicating that scFv are not trapped in the microtubule pellet. To confirm that co-pelleting represents binding to microtubules, GTP was omitted from the assembly mixture to significantly reduced tubulin polymerization. Under these conditions 2F12 and 2G4 are found in the supernatant fraction. B) To confirm that the fraction of scFv 2G4 present in the supernatant was active, we stabilized polymerized microtubules with Taxol in order to obtain a higher ratio of microtubules to soluble tubulin. The same experiment described in (A) was repeated in the absence (labeled “No Tub”), and in the presence of about equal concentrations of microtubules and tubulin (labeled “MT/Tub∼1”) and most tubulin polymerized into microtubules (labeled “MT/Tub>1”). The bands were quantified and the percentage of soluble tubulin, microtubules, and soluble and pelleted scFv are indicated below the lanes. Because of saturation of the Coomassie signal, tubulin quantitations are approximate and the percentage of tubulin in the pellet fraction is underestimated in the last lane.
Non-dynamic microtubules often comprise only a small subset of the total microtubule population within rapidly dividing cells grown in culture
(A) LLCPK cells were fixed 24 h after transfection and stained with antibodies specific for de-tyrosinated α-tubulin. The area bracketed in white is enlarged in (A′). 2G4-GFP does not show detectable binding to microtubules recognized by an antibody specific for detyrosinated α-tubulin. (B) LLCPK cells expressing 2G4-GFP were incubated in 33 µM nocodazole for 15 m prior to fixation and localization of detyrosinated α-tubulin. Depolymerization of the majority of microtubules shifted 2G4-GFP to a soluble protein present uniformly throughout the cell and it did not colocalize with microtubules composed of detyrosinated α-tubulin. Scale bars = 10 µm (whole cell images) and 5 µm (enlarged region). Images shown are maximum intensity projections from optical sections.
Microtubules marked by tubulin acetylation were also examined. Acetylated microtubules were typically more abundant than detyrosinated microtubules in LLCPK cells and the 2G4-GFP signal showed overlap with a subset of the microtubules recognized by anti-acetylated α-tubulin in fixed LLCPK cells (
Based on the lack of 2G4-GFP binding to detyrosinated-a-tubulins in LLCPK cells, we asked whether 2G4 recognizes α- or ß-tubulin isolated from pig brain and whether a-tubulin’s C terminal tyrosine was necessary for scFv binding. By immunoblot, purified His-tagged 2G4 recognized porcine brain α-tubulin, with only minimal recognition of ß-tubulin (
(A) Immunoblot of pig brain extracts probed with scFv’s or anti-tubulin antibodies. Results are shown for the purified recombinant antibodies 13R4 (irrelevant anti-ß-galactosidase scFv), 2G4 or 2F12 as described in Methods. Signals from commercial anti-tubulin polyclonal and monoclonal anti-ß-tubulin antibodies are shown for comparison. (B) Total E. coli extracts from cells expressing GST-tubulin fusion proteins were probed with the scFv 2G4 (10 µg/ml), anti-detyrosinated tubulin, or with an anti-GST antibody (loading control). The terminal tyrosine of α-tubulin is deleted from the delta Y451 fusion protein. See methods section for sequence accession numbers.
Here we described an scFv, 2G4, tagged with GFP, that is soluble when expressed in mammalian cells, colocalizes with microtubules and binds microtubules with an affinity similar to that measured for several microtubule-associated proteins. We posit that the weak binding affinity is sufficient to allow localization, without disrupting microtubule function, as measured here by microtubule assembly dynamics and the binding of EB1, TOGp and CLIP-170 to microtubule ends. For study of the microtubule cytoskeleton, expression of 2G4-GFP provides an alternative to expression of GFP-tagged α-tubulin. Although all evidence to date indicates that the GFP tag does not compromise α-tubulin function, there may be situations where an alternative tag that is not assembled as part of the microtubule could be advantageous. For example, the GFP at the N-terminus of α-tubulin should be positioned within the lumen of the microtubule where it could possibly interfere with studies of tubulin acetylation, a modification localized to the inside surface of the hollow microtubule
Typical screens have focused on identifying high affinity antibodies or scFv (e.g.
The other interesting feature of the 2G4 scFv is that it is able to differentiate between α-tubulins differing only in the presence or absence of the C-terminal tyrosine residue. 2G4 colocalizes with, and binds to, tyrosinated α-tubulin, but not detyrosinated α-tubulin (
In summary, here we describe use of a recombinant antibody as an intracellular tracer, specific for one form of a protein and able to differentiate between unmodified and post-translational protein modifications. Intracellular expression of fluorescently tagged scFv, combined with treatments to enrich for a post-translational modification, should allow isolation of biosensors able to detect specific protein post-translational modifications in living cells.
(TIF)
(TIF)
(M4V)
(M4V)
(M4V)
(M4V)
We are grateful to Nicole Bec for performing tubulin purifications and microtubule pelleting experiments, Mike Matrone and Stuart Martin, University of Maryland, for providing the GFP-tau plasmid, Ned Lamb (IGH, Montpellier, France) for providing rabbit polyclonal serum raised against purified tubulin, Juliette van Dijk for rabbit polyclonal serum raised against detyrosinated a-tubulin, Holly Goodson (Notre Dame University) for providing antibodies to CLIP-170, and Mike Davidson, Florida State University, for providing the mCherry-EB3 plasmid. The Pennsylvania Department of Health specifically disclaims responsibility for any analyses, interpretations, or conclusions.